Both simulation and sequencing data reveal coinfections with multiple SARS-CoV-2 variants in the COVID-19 pandemic

Figures

• 

  Fig. 1. Workflow of SNP clique analysis and the maximal clique in the GISAID20May11 dataset. A. The workflow of SNP clique analysis. First, we constructed an SNP, single-nucleotide polymorphism, and coexisting network from the SNP matrix. Every SNP locus is a vertex, and we add an edge between a locus pair if they have all four major genotypes. We then extract the maximal clique from the network. B. The maximal 16-SNP clique was found in the GISAID20May11 dataset with 11,179 SARS-CoV-2 genomes.

• 

  Fig. 2. The minimum mutation frequency to access n-SNP clique from the reference genome. A, B, and C illustrated the formation of 2, 3, and n-SNP cliques with minimum mutation. Each block represents the alleles at the SNP loci. The yellow blocks (shown with “R”) and the red blocks (shown with “A”) represent major and minor alleles, respectively. The circles with numbers exhibit the loci of the alleles in the reference genome. n represents the size of the maximal clique of the SNP coexistence network. Kn stands for the number of mutations required to form n-clique from the reference genome. From the plot, we concluded that the reference genome experienced at least 2n − 1 mutations to obtain an n-SNP clique. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 3. The simulation flowchart of viral SNVs in samples and the regression of variant number and the coinfection index. A. We simulated the transmission route based on known epidemiological information of SARS-CoV-2 and constructed a transmission tree. Then, we selected the sequenced samples based on their release date in the GISAID database. B. Variant numbers in all samples fit the Poisson distribution with λ equal to the average variant number. In a single transmission branch, variants in child nodes are randomly inherited from the parent sample. In addition, new SNVs, single-nucleotide variants, would also emerge in the child nodes based on the given mutation rate. C. We simulated two possible methods to obtain assembled genomes in the datasets. In method 1, the assembled genome exhibited one of the variants. In method 2, the assembled genome represented the mixed genome of multiple variants. Then, we acquired the SNV list for all samples as the output. D. The distribution of the coinfection index with different average variant numbers in the GISAID20May11 dataset. The dash lines suggested the coinfected variant number corresponding to the 16-SNP clique with method 2.

• 

  Fig. 4. The distribution of samples with different SNVs in the GISAID20May11 dataset and the simulation under different mutation rates. We had 15 possible average numbers of variants and ten duplicates for each pair of mutation rates and the average variant number. We plotted the sample number in all simulations and regressed the sample number vs the number of SNVs of all simulations with a specific mutation rate, and the 95% CI region is shown in grey.

• 

  Fig. 5. The maximal SNP clique and the coinfection analysis in the GISAID21Apr1 dataset. A. The maximal SNP clique in the GISAID21Apr1 dataset. The 4113 SARS-CoV-2 genomes in the GISAID21Apr1 dataset contained the maximal SNP clique of size 34. B. The regression of variant number and the coinfection index in the GISAID21Apr1 dataset. The dash line exhibited that the averaged variant number of 3.42 corresponding to the coinfection index of 34 in the GISAID21Apr1 dataset.

• 

  Fig. 6. Statistics of Nanopore sequencing data and the distribution of recurrent SNPs for the 42 COVID-19 samples. A. After low-quality filtration, we aligned the Nanopore sequencing reads to the SARS-CoV-2 genome and human transcriptome. The histograms in red and blue exhibit the number of reads aligned to the SARS-CoV-2 genome and human transcriptome, respectively. B. The heatmap exhibited twenty recurrent SNPs for the 42 COVID-19 patients. The squares with deeper colour represent the higher ratios of the alternate allele in the sample. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 7. The procedure of chimeric read identification and read splicing. To remove the chimeric reads, we first aligned the multiplex PCR primer to the Nanopore sequencing reads and split the reads into segments according to the aligning results. Then, we aligned the read segments to the SARS-CoV-2 genome and discarded those segments that could not be aligned to the reference genome. Finally, we removed the original reads and kept the genome-aligned segments.

• 

  Fig. 8. Coverage of depth of aligned data in the 42 COVID-19 samples. The X coordinate represents the location of the SARS-CoV-2 genome, and the Y coordinate represents the sequencing depth. The red, yellow, green, pink, brown, light green, purple and dark brown bars represent the genetic regions of ORF1ab, S, ORF3a, M, ORF6, ORF8, N and ORF10, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 9. Distributions of heterozygous SNPs in the 42 COVID-19 samples. After obtaining the SARS-CoV-2 genomic reads, we detected the SNPs for each sample and exhibited the genotypes for the reference and mutations. For each sample, the mutated and reference genotypes were marked with red and blue, respectively. At the top of the figure, we exhibited the locations of the genes in the SARS-CoV-2 genome and annotated them with different colours. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)